Learning deep control policies for autonomous aerial vehicles with MPC-guided policy search

Zhang, Tianhao; Kahn, Gregory; Levine, Sergey; Abbeel, Pieter

doi:10.1109/icra.2016.7487175

Cited by 353 publications

(248 citation statements)

References 31 publications

Supporting

Mentioning

232

Contrasting

Unclassified

Order By: Relevance

“…• Robotics DNNs have been successful in the domain of robotic tasks such as grasping with a robotic arm [32], motion planning for ground robots [33], visual navigation [4,34], control to stabilize a quadcopter [35] and driving strategies for autonomous vehicles [36]. DNNs are already widely used in multimedia applications today (e.g., computer vision, speech recognition).…”

Section: E Applications Of Dnnmentioning

confidence: 99%

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

et al. 2017

View full text Add to dashboard Cite

Abstract-Deep neural networks (DNNs) are currently widely used for many artificial intelligence (AI) applications including computer vision, speech recognition, and robotics. While DNNs deliver state-of-the-art accuracy on many AI tasks, it comes at the cost of high computational complexity. Accordingly, techniques that enable efficient processing of DNNs to improve energy efficiency and throughput without sacrificing application accuracy or increasing hardware cost are critical to the wide deployment of DNNs in AI systems.This article aims to provide a comprehensive tutorial and survey about the recent advances towards the goal of enabling efficient processing of DNNs. Specifically, it will provide an overview of DNNs, discuss various hardware platforms and architectures that support DNNs, and highlight key trends in reducing the computation cost of DNNs either solely via hardware design changes or via joint hardware design and DNN algorithm changes. It will also summarize various development resources that enable researchers and practitioners to quickly get started in this field, and highlight important benchmarking metrics and design considerations that should be used for evaluating the rapidly growing number of DNN hardware designs, optionally including algorithmic co-designs, being proposed in academia and industry.The reader will take away the following concepts from this article: understand the key design considerations for DNNs; be able to evaluate different DNN hardware implementations with benchmarks and comparison metrics; understand the trade-offs between various hardware architectures and platforms; be able to evaluate the utility of various DNN design techniques for efficient processing; and understand recent implementation trends and opportunities.

show abstract

Section: E Applications Of Dnnmentioning

confidence: 99%

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

et al. 2017

View full text Add to dashboard Cite

show abstract

“…The work of Zhang et al [125] highlights two additional challenges. First, unsupervised learning is not practical for robotic systems where a single failure is catastrophic, as in aerial vehicles.…”

Section: Current Shortcomings Of Dnns For Roboticsmentioning

confidence: 99%

“…They varied video playback speed to simulate this temporal variance, expanding their training set without the need to acquire additional samples. Still other researchers utilizing reinforcement learning, such as Polydoros et al [115] and Zhang et al [125], automated training using alternative control systems during the learning phase.…”

Section: Current Shortcomings Of Dnns For Roboticsmentioning

confidence: 99%

“…The canonical model for using deep neural networks for learning a control policy is deep Q-learning [47]. It Ultimately, reinforcement learning models are useful for learning to operate dynamic systems from partial state information, and controllers based on deep reinforcement learning can be very computationally efficient at runtime [125]. They automatically infer priorities based on rewards that are obtained during training.…”

Section: The Role Of Structure D In Roboticsmentioning

confidence: 99%

“…Zhang, Kahn, and Levine [125] learned a control policy to implement a model predictive control guided search for autonomous aerial vehicles. Reducing computational load in mobile robotics translates into power savings that increase range and/or improve performance.…”

Section: Examples In Recent Researchmentioning

confidence: 99%

See 2 more Smart Citations

Deep learning in robotics: a review of recent research

2017

View full text Add to dashboard Cite

Advances in deep learning over the last decade have led to a flurry of research in the application of deep artificial neural networks to robotic systems, with at least thirty papers published on the subject between 2014 and the present. This review discusses the applications, benefits, and limitations of deep learning vis-à-vis physical robotic systems, using contemporary research as exemplars. It is intended to communicate recent advances to the wider robotics community and inspire additional interest in and application of deep learning in robotics.

show abstract

Nonlinear and Adaptive Intelligent Control Techniques for Quadrotor UAV – A Survey

Mo¹,

Farid²

2018

Asian Journal of Control

165

View full text Add to dashboard Cite

Parametric uncertainties and coupled nonlinear dynamics are inherent in quadrotor configuration and infer adaptive nonlinear approaches to be used for flight control system. Numerous adaptive nonlinear and intelligent control techniques, which have been reported in the literature for designing quadrotor flight controller by various researchers, are investigated in this paper. As a priori, each conventional nonlinear control technique is discussed broadly and then its adaptive/observer based augmentation is conferred along with all possible variants. Among conventional nonlinear control approaches, feedback linearization, backstepping, sliding mode, and model predictive control, are studied. Intelligent control approaches incorporating fuzzy logic and neural networks are also discussed. In addition to adaption based parametric uncertainty handling, various other aspects of each control technique regarding stability, disturbance rejection, response time, asymptotic, exponential and finite time convergence etc., are discussed in sufficient depth. The contribution of this paper is the provision of detailed and in depth discussion on quadrotor nonlinear control approaches to the flight control designers.

show abstract

Learning deep control policies for autonomous aerial vehicles with MPC-guided policy search

Cited by 353 publications

References 31 publications

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

Deep learning in robotics: a review of recent research

Nonlinear and Adaptive Intelligent Control Techniques for Quadrotor UAV – A Survey

Contact Info

Product

Resources

About